Virginia Tech® home

Nuclear Science Week: Building Machine Learning for Hazards

Nuclear Science Week
Duarte with students
Duarte's lab members in a photo taken before the COVID-19 pandemic. Front row, left to right: Bruno Serrao, Paul Hurley, Juliana Duarte, Abdulsalam Shakhatreh. Back row, left to right: Elvan Sahin, Jeric Demasana, Abdulla Alblooshi, Nick Burns.

Nuclear power plants supply vast amounts of energy to the communities they serve. Since the first plant was deployed in the Soviet Union in 1954, understanding of this method of energy production as well as building of plants has steadily increased. The U.S. built its first plant in 1958, and today is home to 58 plants with 96 commercial reactors. Today, nuclear power supplies a fifth of American electricity.

Nuclear energy requires less refueling and less maintenance than other types of power plants, making this energy source by far the most reliable. To maintain this efficiency, it is important to maintain safety measures for both the plant and its personnel. These measures are one of the major emphases of the nuclear engineering program at Virginia Tech. Researcher Juliana Pacheo Duarte leads a team that is working to create machine learning to predict hazards such as natural disasters and fires that could have an effect on the facilities, and and $800,000 grant from the Department of Energy is carrying the effort further.